GRET: Global Representation Enhanced Transformer

We introduce the discussion mechanism into the multiagent communicating encoder-decoder architecture for Natural Language Generation (NLG) tasks and prove that by applying the discussion mechanism, the communication between agents becomes more effective. Generally speaking, an encoder-decoder architecture predicts target-sequence word by word in several time steps. At each time step of prediction, agents with the discussion mechanism predict the target word after several discussion steps. In the first step of discussion, agents make their choice independently and express their decision to other agents. In the next discussion step, agents collect other agents’ decision to update their own decisions, then express the updated decisions to others again. After several iterations, the agents make their final decision based on a well-communicated situation. The benefit of the discussion mechanism is that multiple encoders can be designed as different structures to fit the specified input or to fetch different representations of inputs.We train and evaluate the discussion mechanism on Table to Text Generation, Text Summarization and Image Caption tasks, respectively. Our empirical results demonstrate that the proposed multi-agent discussion mechanism is helpful for maximizing the utility of the communication between agents.

The advances in deep learning have led to great achievements in many Natural Language Processing (NLP) tasks. With the nature of language, i.e., sequential data, most NLP tasks can be framed into the sequence learning framework, such as text generation. As one of the most important foundations for modern NLP techniques, autoregressive generation models have achieved dominant performance in a great deal of NLP tasks. Therefore, this thesis emphasizes improving the autoregressive generation model for different NLP tasks. While many tasks can naturally fit into the sequence learning framework, some of them, e.g., building discourse parsing tree, require sophisticated designs to fit into neural models. Therefore, this thesis firstly emphasizes a novel unified framework for discourse parsing, which builds a discourse tree in a top-down depth-first manner, and it frames the task as an autoregressive generation task with the goal of each step being the prediction of the node position given a piece of text. The proposed approach is proven effective with extensive empirical experiments. In addition, I extend the above framework by proposing a hierarchical decoder, which leverages the information from parents and siblings of the nodes that are currently processed. The proposed decoder utilizes the nature of the tree structure and further improves the experiment performance on both discourse parsing and dependency parsing tasks. On the other hand, the de facto strategies, i.e., cross entropy loss and teacher forcing, for training the autoregressive generation models have been shown problematic in certain aspects. For example, cross entropy loss, which is one of the widely leveraged training objective functions, often leads to text degeneration in text generation, and teacher forcing suffers from the exposure bias problem, where there exists a mismatch between the training and testing setup. For text degeneration, I introduce a class of diminishing attentions, which enforces the submodularity of the coverage calculated by cross attention in the sequence-to-sequence model. The proposed diminishing attentions achieve notable improvement on several neural text generation tasks, including text summarization, machine translation, and image paragraph generation. Further, I propose a novel training objective, ScaleGrad, to replace cross entropy, which significantly reduces the degeneration problem in different text generation tasks. In fact, ScaleGrad can be extended to problems beyond text degeneration. It provides wide flexibility to inject different inductive biases into the text generation model by directly modifying the gradient information in the output layer. Next, for the exposure bias problem, this thesis introduces a novel type of scheduled sampling based on training accuracy, which requires only minimal hyper-parameter tuning compared to existing scheduled sampling methods. Additionally, a novel imitation loss is proposed to further enforce the model’s generative behavior to match the teacher-forced behavior. Moreover, this thesis demonstrates that reducing exposure bias can improve the robustness of language models against repetition and toxic errors.

text summarization is a classic sequence-to-sequence natural language generation task. In order to improve the quality of unsupervised abstract text summarization in unsupervised mode, we propose two constraints for training text summarization model, embedding space constraint and information ratio constraint. We construct a generative adversarial network with two discriminators based on these two constraints (TC-SUM-GAN). We use unsupervised and supervised methods to train the model in the experiment. Experimental results show that the ROUGE-1 value of the unsupervised TC-SUM-GAN increases by [Formula: see text] points compared with the basic model and at least 1.96 points compared with other comparative models. The ROUGE scores of the supervised TC-SUM-GAN are also improved. TC-SUM-GAN achieves very competitive results for the metrics of ROUGE-1 and ROUGE-2. In addition, the abstracts generated by our model are closer to those generated manually.

This paper addresses the problem of automatic generation of natural language descriptions for ontology-described artifacts. The original motivation for the work is the challenge of providing textual narratives of automatically generated scientific workflows (e.g., paragraphs that scientists can include in their publications). The paper presents two systems which generate descriptions of sets of atoms derived from a collection of ontologies. The first system, called nlgPhylogeny, demonstrates the feasibility of the task in the Phylotastic project, providing evolutionary biologists with narrative for automatically generated analysis workflows. nlgPhylogeny utilizes the fact that the Grammatical Framework (GF) is suitable for the natural language generation (NLG) task; the paper shows how elements of the ontologies in Phylotastic, such as web services and information artifacts, can be encoded in GF for the NLG task. The second system, called \(\mathtt{\small nlgOntology}^{A}\), is a generalization of nlgPhylogeny. It eliminates the requirement that a GF needs to be defined and proposes the use of annotated ontologies for NLG. Given a set of annotated ontologies, \(\mathtt{\small nlgOntology}^{A}\) generates a GF suitable for the creation of natural language descriptions of sets of atoms derived from these ontologies. The paper describes the algorithms used in the development of nlgPhylogeny and \(\mathtt{\small nlgOntology}^{A}\) and discusses potential applications of these systems.

Despite the success of text-to-text pre-trained models in various natural language generation (NLG) tasks, the generation performance is largely restricted by the number of labeled data in downstream tasks, particularly in data-to-text generation tasks. Existing works mostly utilize abundant unlabeled structured data to conduct unsupervised pre-training for task adaption, which fail to model the complex relationship between source structured data and target texts. Thus, we introduce self-training as a better few-shot learner than task-adaptive pre-training, which explicitly captures this relationship via pseudo-labeled data generated by the pre-trained model. To alleviate the side-effect of low-quality pseudo-labeled data during self-training, we propose a novel method called Curriculum-Based Self-Training (CBST) to effectively leverage unlabeled data in a rearranged order determined by the difficulty of text generation. Experimental results show that our method can outperform fine-tuning and task-adaptive pre-training methods, and achieve state-of-the-art performance in the few-shot setting of data-to-text generation.

Despite the recent advancement in NLP research, cross-lingual transfer for natural language generation is relatively understudied. In this work, we transfer supervision from high resource language (HRL) to multiple low-resource languages (LRLs) for natural language generation (NLG). We consider four NLG tasks (text summarization, question generation, news headline generation, and distractor generation) and three syntactically diverse languages, i.e., English, Hindi, and Japanese. We propose an unsupervised cross-lingual language generation framework (called ZmBART) that does not use any parallel or pseudo-parallel/back-translated data. In this framework, we further pre-train mBART sequence-to-sequence denoising auto-encoder model with an auxiliary task using monolingual data of three languages. The objective function of the auxiliary task is close to the target tasks which enriches the multi-lingual latent representation of mBART and provides good initialization for target tasks. Then, this model is fine-tuned with task-specific supervised English data and directly evaluated with low-resource languages in the Zero-shot setting. To overcome catastrophic forgetting and spurious correlation issues, we applied freezing model component and data argumentation approaches respectively. This simple modeling approach gave us promising results.We experimented with few-shot training (with 1000 supervised data points) which boosted the model performance further. We performed several ablations and cross-lingual transferability analyses to demonstrate the robustness of ZmBART.

Chinese medical named entity recognition (CMNER) aims to extract entities from Chinese unstructured medical texts. Existing character-based NER models do not comprehensively consider character’s characteristics from different perspectives, which limits their performance in applying to CMNER. In this paper, we propose a local and global character representation enhanced model for CMNER. For the input sentence, the model fuses the spacial and sequential character representation using autoencoder to get the local character representation; extracts the global character representation according to the corresponding domain words; integrates the local and global representation through gating mechanism to obtain the enhanced character representation, which has better ability to perceive medical entities. Finally, the model sent the enhanced character representation to the Bi-LSTM and CRF layers for context encoding and tags decoding respectively. The experimental results demonstrate that our model achieves a significant improvement over the best baseline, increasing the F1 values by 1.04% and 0.62% on the IMCS21 and CMeEE datasets, respectively. In addition, we verify the effectiveness of each component of our model by ablation experiments.

The task of natural language generation by computer requires a method for the selection of lexical items that refer to concepts. A generation system cannot assume that every concept will have an appropriate lexical entry associated with it. Similarly, a generation system must avoid redundant text and must strive for the most specific descriptions possible. This paper presents an algorithm for lexical selection that takes advantage of a conceptual network in which we represent the relationships between concepts (such as which concepts subsume each other and what differentiates concepts at different levels of generality). When the algorithm chooses a lexical realization for a concept, it first checks to see if the concept to be expressed is associated with a lexical entry. If so, that entry is used. If the concept does not have the appropriate lexical associations, the algorithm generates a phrase with a more general head term and restrictive modifiers. It does this by using the conceptual network to compute a semantically appropriate head term, and then modifying the request for generation by adding restrictions that differentiate the concept associated with the head term from the concept initially requested. Similarly, the conceptual network is used to eliminate redundant modifiers by computing the information contained in a lexical item that is chosen as a head term. Here the algorithm modifies the generation request to avoid restating the extra information. This algorithm is being implemented within Penman, a computerized system for the generation of English text from statements in a first-order predicate calculus language.

In this paper, we propose Sequence Span Rewriting (SSR), a self-supervised task for sequence-to-sequence (Seq2Seq) pre-training. SSR learns to refine the machine-generated imperfect text spans into ground truth text. SSR provides more fine-grained and informa- tive supervision in addition to the original text- infilling objective. Compared to the preva- lent text infilling objectives for Seq2Seq pre- training, SSR is naturally more consistent with many downstream generation tasks that re- quire sentence rewriting (e.g., text summariza- tion, question generation, grammatical error correction, and paraphrase generation). We conduct extensive experiments by using SSR to improve the typical Seq2Seq pre-trained model T5 in a continual pre-training setting and show substantial improvements over T5 on various natural language generation tasks.

Natural language generation (NLG) spans a broad range of tasks, each of which serves for specific objectives and desires different properties of generated text. The complexity makes automatic evaluation of NLG particularly challenging. Previous work has typically focused on a single task and developed individual evaluation metrics based on specific intuitions. In this paper, we propose a unifying perspective based on the nature of information change in NLG tasks, including compression (e.g., summarization), transduction (e.g., text rewriting), and creation (e.g., dialog). Information alignment between input, context, and output text plays a common central role in characterizing the generation. With automatic alignment prediction models, we develop a family of interpretable metrics that are suitable for evaluating key aspects of different NLG tasks, often without need of gold reference data. Experiments show the uniformly designed metrics achieve stronger or comparable correlations with human judgement compared to state-of-the-art metrics in each of diverse tasks, including text summarization, style transfer, and knowledge-grounded dialog.

This chapter discusses the concept of natural language processing (NLP), specifically how the NLP is applied. The chapter further outlines the typical functions of the NLP. In addition, it explores the NLP in strategy and strategy implementation.The NLP is a branch of computer science, AI and linguistics that focuses on the interaction between computer programs and human language. There are three functions of the NLP, namely, machine translation (MT), text summarisation and sentiment analysis (SA). MT is a branch of computer linguistics that analyses the use of computerised tools to translate the context derived from this, from one language to the next. The language that is being translated is the human language.Automatic text summarisation refers to a process of bringing about a succinct and meaningful summary of the text. SA is a way of finding out the polarity or strength of the opinion that is expressed in written text. The opinion could be either positive or negative. MT has three major approaches, namely, rule-based MT, statistical MT and neural MT. A rule-based system would typically require the expert to know both the source and the target languages. This knowledge is then used to develop syntactic, semantic and morphological rules to achieve the translation. The statistical approach uses statistical models based on the analysis of bilingual text corpora. The neural MT approach uses neural networks to achieve machine translation. There are two broad categories of text summarisation approaches, namely, extractive summarisation and abstractive summarisation. Furthermore, there are four popular types of SA, namely, the fine-grained SA, emotion detection, aspect-based SA and multilingual SA. The firm that deploys the NLP and ML in Big Data Analytics would gain crucial and better insights, setting the firm apart from its competitors.

Speech recognition and text summarization, plagiarism detection, machine translation, chatbots, sentiment analysis (SA), question answering (QA), and dialogue systems are all products of natural language processing (NLP), a branch of AI concerned with modeling natural languages for the purpose of developing relevant applications. NLP draws on several disciplines, not only computer science and linguistics, for its research and development. These include cognitive science, psychology, mathematics, and more. More than 1.5 billion Muslims throughout the world depend on the Arabic language for their daily five-times-prayer practice, and Arabic is one of six official languages used by more than 422 million people in the Arab world, according to UNESCO. Dialectal Arabic is the slang language spoken informally in everyday life and varies from country to country; Classical Arabic is a reflection of the language spoken by the Arabs more than fourteen centuries ago. Modern Standard Arabic is an evolving variety of Arabic that borrows and innovates regularly to meet the changing needs of its speakers. Complexity is added to the Arabic language by the fact that it encompasses not one but three different varieties of spoken language: classical, contemporary, and colloquial. However, Arabic language processing on computers remains difficult for a variety of reasons. Because of the extensive inflection and derivational processes that occur in Arabic, a single lemma may be utilized to produce several words with distinct meanings.

Machine Translation (MT) which has been championed as an effective technology for knowledge transfer from English to languages with less digital content. An example of such efforts is the automatic translation of English Wikipedia to languages with smaller collections, such as Arabic. However, MT quality is still far from ideal for many of the languages and text genres. While translating a document, many sentences are poorly translated which can provide an incorrect text, and confuse the reader. Moreover, some of these sentences are not as informative and could be summarized to make a more cohesive document. Thus, for tasks in which complete translation is not mandatory, MT can be effective if the system can provide an informative subset of the content with higher translation quality. For this scenario, text summarization can provide effective support for MT by keeping only the most important and informative parts of a given document to translate. In this work, we demonstrate a framework of MT and text summarization which replaces the baseline translation with a proper summary that has higher translation quality than the full translation. For this, we combine the state of the art English summarization system and a novel framework for prediction of MT quality without references. Our framework is composed of the following major components: (a) a standard machine translation system, (b) a reference-free MT quality estimation system, (c) an MT-aware summarization system, and (d) an English-Arabic sentence matcher. More specifically, our English-Arabic system reads in an English document along with its baseline Arabic translation and outputs, as a summary, a subset of the Arabic sentences based on their informativeness and their translation quality. We demonstrate the utility of our system by evaluating it with respect to its translation and summarization quality and demonstrate that we can balance between improving MT quality and maintaining a decent summarization quality. For summarization, we conduct both reference-based and reference-free evaluations and observe a performance in the range of the state of the art system. Moreover, the translation quality of the summaries shows an important improvement against the baseline translation of the entire documents. This MT-aware summarization approach can be applied to translation of texts such as Wikipedia articles. For such domain-rich articles, there is a large variation of translation quality across different sections. An intelligent reduction of the translation tasks results in improved final outcome. Finally, the framework is mostly language independent and can be easily customized for different target languages and domains.

Due to the exponential overflow of textual information in various fields of knowledge and on the internet, it is very challenging to extract important information or to generate a summary from some multi-document collection in a specific field. With such a gigantic amount of textual content, human text summarization becomes impractical since it is expensive and consumes a lot of time and effort. So, developing automatic text summarization (ATS) systems is becoming increasingly essential. ATS approaches are either extractive or abstractive. The extractive approach is simpler and faster than the abstractive approach. This work proposes an extractive ATS system that aims to extract a small subset of sentences from a large multi-document text. First, the whole text is preprocessed by applying some natural language processing techniques such as sentences segmentation, words tokenization, removal of stop-words, and stemming to provide a structured representation of the original document collection. Based on this structured representation, the ATS problem is formulated as a multi-objective optimization (MOO) problem that optimizes the extracted summary to maintain the coverage of the main text content while avoiding redundant information. Secondly, an evolutionary sparse multi-objective algorithm is developed to solve the formulated large-scale MOO. The output of this algorithm is a set of non-dominated summaries (Pareto front). A novel criterion is proposed to select the target summary from the Pareto front. The proposed ATS system has been examined using (DUC) datasets, and the output summaries have been evaluated using (ROUGE) metrics and compared with the literature.

Large language models (LLMs), consisting of billions and trillions of parameters, have demonstrated exceptional ability in natural language understanding (NLU) and natural language generation (NLG) tasks. Increases in their numbers of parameters and model sizes have resulted in better performance and accuracy. However, models with such enormous numbers of parameters incur significant computational costs and resources, making them challenging to fine tune and adapt to a specific downstream task. Several parameter-efficient fine-tuning (PEFT) techniques have been proposed to address this issue. This study demonstrates the improvement obtained over the base LLaMA3-8B model using two prominent PEFT techniques: LoRA and QLoRA. We use the sequence classification task of sentiment analysis to conduct the experiments. Additionally, we analyze the effects of hyperparameter adjustments (r and α) on the model’s performance. We examine the tradeoff between efficiency and memory savings obtained using the quantized LoRA (QLoRA) technique. We also investigate and compare the performance changes of LoRA and QLoRA techniques obtained after adapting to attention layers (query, key, value, and project) to all the linear layers during fine tuning. We report the findings of our work along with limitations and future directions.